A Network Traffic Classification Using C5.0 Algorithm

Abstract

In this globalized era, the present scenario involves a high speed internet infrastructure. So the monitoring of the network traffic becomes a tedious task. Its analysis requires study and knowledge about various types of applications which form the network traffic. Various methods like Deep Packet Inspection (DPI), Machine Learning Algorithms (MLA's) have in practice over years for analyzing and classification of network traffic. This paper presents the concept of network traffic classification using C5.0 algorithm which is the latest algorithm out of the machine learning algorithms. It uses a C5.0 classifier and while capturing packets our machine learns itself by machine learning algorithm. A level of accuracy is obtained by using high quality of training data, a unique set of parameters are taken into account for both training and classification. This type of comparison is preferable over others as it is able to distinguish among seven different applications in a test set of range of seventy thousand to one lakh unknown cases with an average accuracy of 99.3-99.9%.